Robust Noise Estimation Based on Noise Injection

  • Authors:

  • Chongwu Tang;Xiaokang Yang;Guangtao Zhai

  • Affiliations:

  • Shanghai Key Labs of Digital Media Processing and Communication, Shanghai Jiao Tong University, Shanghai, China;Shanghai Key Labs of Digital Media Processing and Communication, Shanghai Jiao Tong University, Shanghai, China;Shanghai Key Labs of Digital Media Processing and Communication, Shanghai Jiao Tong University, Shanghai, China

  • Venue:
  • Journal of Signal Processing Systems
  • Year:
  • 2014

Quantified Score

Hi-index 0.00

Visualization

Abstract

Noise estimation is an important premise for image denoising and many other image processing applications, and related research has drawn increasing attention and interest. In this paper, a novel noise level estimation algorithm is proposed by investigating the distribution of local variances in natural images. There are two major contributions of this work to tackle with the challenges in noise estimation: 1) a wavelet decomposition based preliminary estimation stage to alleviate the influence of image's textural or structural information; 2) a noise injection based estimation stage to simulate the impact of noise-free image content on the variance distribution, which otherwise almost always leads to overestimation. The cascade scheme of this two-step estimation procedure can reduce the detrimental influence of textural image regions effectively and therefore relieves overestimation of the noise variance. Moreover, the proposed method is not limited to any specific type of noise distribution. Extensive experiments and comparative analysis demonstrate that the proposed algorithm can reliably infer noise levels and has robust performance over a wide range of visual content, as compared to relevant methods.